ZEUS 8B 🌩️ V24

This is a merge of pre-trained language models created using mergekit.

Merge Details

Merge Method

This model was merged using the DARE TIES merge method using unsloth/Meta-Llama-3.1-8B-Instruct as a base.

Models Merged

The following models were included in the merge:

Configuration

The following YAML configuration was used to produce this model:

base_model: Skywork/Skywork-o1-Open-Llama-3.1-8B
dtype: bfloat16
merge_method: slerp
parameters:
  t:
  - value: 0.5
slices:
- sources:
  - layer_range: [0, 32]
    model: Skywork/Skywork-o1-Open-Llama-3.1-8B
  - layer_range: [0, 32]
    model: FreedomIntelligence/HuatuoGPT-o1-8B
---
base_model: unsloth/Meta-Llama-3.1-8B-Instruct
dtype: bfloat16
merge_method: dare_ties
parameters:
  int8_mask: 1.0
  normalize: 1.0
  random_seed: 145.0
slices:
- sources:
  - layer_range: [0, 32]
    model: unsloth/Llama-3.1-Storm-8B
    parameters:
      density: 0.94
      weight: 0.35
  - layer_range: [0, 32]
    model: arcee-ai/Llama-3.1-SuperNova-Lite
    parameters:
      density: 0.92
      weight: 0.26
  - layer_range: [0, 32]
    model: VAGOsolutions/Llama-3.1-SauerkrautLM-8b-Instruct
    parameters:
      density: 0.91
      weight: 0.2
  - layer_range: [0, 32]
    model: Orenguteng/Llama-3.1-8B-Lexi-Uncensored-V2
    parameters:
      density: 0.93
      weight: 0.19
  - layer_range: [0, 32]
    model: output\strawberry-patch
    parameters:
      density: 0.92
      weight:
      - filter: self_attn.o_proj
        value: 0.26
      - filter: mlp.down_proj
        value: 0.26
      - filter: layers.18.
        value: 0.26
      - filter: layers.19.
        value: 0.26
      - filter: layers.20.
        value: 0.26
      - value: 0.0
  - layer_range: [0, 32]
    model: unsloth/Meta-Llama-3.1-8B-Instruct
tokenizer:
  tokens:
    <|begin_of_text|>:
      force: true
      source: unsloth/Meta-Llama-3.1-8B-Instruct
    <|eot_id|>:
      force: true
      source: unsloth/Meta-Llama-3.1-8B-Instruct
    <|finetune_right_pad_id|>:
      force: true
      source: unsloth/Meta-Llama-3.1-8B-Instruct
    pad_token:
      force: true
      source:
        kind: model_token
        model: unsloth/Meta-Llama-3.1-8B-Instruct
        token_id: 128004

Open LLM Leaderboard Evaluation Results

Detailed results can be found here! Summarized results can be found here!

Metric	Value (%)
Average	21.95
IFEval (0-Shot)	60.00
BBH (3-Shot)	26.15
MATH Lvl 5 (4-Shot)	13.90
GPQA (0-shot)	1.57
MuSR (0-shot)	4.71
MMLU-PRO (5-shot)	25.38

T145
/

ZEUS-8B-V24